AC_PROG_CC adds -g and -O2 to CFLAGS if you're using GCC; don't bother
adding either of them, or -O, ourselves if we're using GCC.
Add -O for all non-GCC compilers.
Don't assume we have <stdint.h>. Instead, use the AC_TYPE_ macros to
ensure we have the C99 intN_t and uintN_t types; we already include
<inttypes.h> in tcpdump-stdinc.h iff we have it.
Get rid of the structure declarations in openflow-1.0.h, as they have
zero-length arrays (not supported by all the compilers people might be
using) and as
1) they're only used in sizeof()
and
2) after each one there's an assertion to check that sizeof()
returns a specific numerical value
so, instead, just #define various _LEN items to those numerical values
and use them.
Add an openflow.h header with a #define for the length of the basic
header, and move the declaration of of10_header_body_print() there.
Some versions of Sun C support __attribute__ but don't support the
"unused" or "format" attributes - they don't fail, but they whine a lot.
They also support the "noreturn" attribute, but don't allow it to be
applied to function pointers, only to functions.
Check whether they can be used without warnings, and use them only if
they can.
autoconf 2.61 and later do, in AC_PROG_CC, tests to figure out how to
beat non-GCC compilers into doing the closest thing to ANSI C of which
they're capable, so we don't need to do them ourselves.
Glibc 2.14 doesn't install the ONC RPC headers, but it installs the ONC
RPC routines, presumably for binary compatibility. Don't use
getrpcbynumber() unless we have it *and* the header file to declare it.
This also fixes configure-script problems on OSes that have IPv6 support
but don't have <netinet6/in6.h> - no standard I could find requires that
header for IPv6 support (the Single UNIX Specification speaks of
<netinet/in.h> for both IPv4 and IPv6, and RFCs 2292 and 3542 don't
mention <netinet6/in6.h>, just <netinet/in.h> and <netinet6/ip6.h>).
Rather than hardcode the WIN32 define, add proper fork checks to the
configure script and check those. This fixes building for nommu systems
which lack the fork function.
While we're here though, add support for this functionality via vfork
so that it does work on nommu systems. And fix an old bug where we
exit properly in the forked child when the exec failed instead of just
returning to the calling code (which isn't expecting it).
Reviewed-By: Guy Harris <guy@alum.mit.edu>
tcpdump.
Check whether __attribute__((format)) can be applied to function
pointers and, if not, don't apply it to function pointers; some older
versions of GCC appear to support applying it to functions, but not
pointers to functions.
As we've gotten rid of missing/getaddrinfo.c, don't use it.
Remove unused items, and items for which we now have AC_DEFINE() calls
that supply a comment, from acconfig.h.
Note that a host OS that matches "osf*" is assumed to be DEC
OSF1/Digital UNIX/Tru64 UNIX.
printed when trying to print packets for a DLT_ for which we don't have
a printer to indicate that you can still save to a capture file in that
case. (Slightly changed not to require DLT_USB_LINUX_MMAPPED to be
defined, for older libpcaps that only define DLT_USB_LINUX.)
with which it was released, and DLT_BLUETOOTH_HCI_H4_WITH_PHDR might be
defined by pcap-bpf.h without pcap/bluetooth.h being present (as appears
to be the case on Fedora 9, for example), so check whether
<pcap/bluetooth.h> is usable.
Update a comment - F9 appears to have a "/usr/include/pcap.h" if you
install the libpcap headers.
When adding -I flags when running a compiler-based test, add them to
CPPFLAGS, not CFLAGS - the latter doesn't work right with
AC_CHECK_HEADERS, as the "gcc -E" run doesn't have the -I flags added.
if it does, use that for the pf definitions;
if it doesn't, don't compile in pf support;
as both OpenBSD and FreeBSD have changed the pf definitions and header
format without changing the DLT value, so you can't reliably read
pflog-format libpcap files on a machine running an OS version other than
the one on which the file was generated.
and using ftell(); that won't necessarily work on Windows (if libpcap
was built with a different version of the C runtime library than tcpdump
is), and, even on UN*X, would break if a pcap_dumper_t * were ever made
something other than a FILE *.
Provide a pcap_dump_ftell() implementation that does that cheating cast
for use if libpcap doesn't have it (a pcap_dumper_t * is just a FILE *
on those older versions of libpcap).
inline problem, and AC_LBL_C_INLINE checks for a case that some versions
of the HP C compiler don't handle, and only uses inline if that case
succeeds.
1. On AIX, AC_LBL_C_INLINE detected the compiler supported
the inline keyword which is wrong. AC_C_INLINE from
autoconf-2.59 worked.
2. AC_CHECK_TYPE from autoconf-2.5x is no longer broken.
Replaced AC_LBL_CHECK_TYPE with it, mainly to use
<sys/bitypes.h> for Tru64 UNIX where some of the u_int#_t
types are defined.
3. Tru64 UNIX 4.0D doesn't support %llx; however, it does support %lx.
4. Added <stdint.h> to interface.h for int#_t types on
Tru64 UNIX 4.0D (required for missing/snprintf.c).
5. Reworked includes in tcpdump-stdinc.h for int#_t types.
do in libpcap for ether_hostton(). Include <netinet/ether.h> only if it
declares ether_ntohost(). If nothing declares it, declare it ourselves,
as we do in libpcap.
Don't cast the second argument to ether_ntohost() to a const pointer, as
some systems don't modify it but don't declare that argument as a const
pointer. (This is similar to what we do on libpvap for
ether_hostton().)
Fix indentation.
causes the configure script to attempt to define the PRI[doux]64 macros
if they're not defined by including <inttypes.h>, and causes
"missing/bittypes.h" to attempt to define them, if undefined, in a
fashion that should, with any luck, work on MSVC++ and various flavors
of GNU C on Windows.
Fix the spelling of "Mac OS X".
causes "int64_t" and "u_int64_t" to be defined by the configure script,
and causes "missing/bittypes.h" to attempt to define "u_int64_t" in a
fashion that should, with any luck, work on MSVC++ and various flavors
of GNU C on Windows.
When saving to a file with "-w", have the "-v" flag make tcpdump
report, every 10 seconds, the number of packets captured.
Include <smi.h> if we're building with libsmi, to declare
"smiInit()" and "smiLoadModule()".
isn't always very suggestive - for example, somebody might think
"EN10MB" is always 10 MB/s, and might not know that "IEEE802" is Token
Ring), using "pcap_datalink_val_to_description()". Supply our own
"pcap_datalink_val_to_description()" if libpcap doesn't have it (even if
it has "pcap_datalink_name_to_val()").
safer way to terminate "pcap_loop()" in a signal handler (it just sets a
flag, it doesn't muck with data structures that might have been in the
middle of being updated).
Have "setsignal()" not request SA_RESTART, so that if we call
"pcap_breakloop()" in a signal handler and then return, we don't restart
a call that was waiting for captured packets, we just make that call
EINTR out.
inet_ntop/inet_pton/inet_aton, and don't use AC_REPLACE_FUNCS to set
LIBOBJS to include the replacement module, as that does the full check
for the routine again, not just the simple replacement.
Move the tests in question after AC_LBL_LIBPCAP, so that we link with
the appropriate libraries (e.g., "-lsocket -lnsl" on Solaris) to use
when using those routines.
the output stream for "-w" to be flushed after each packet is dumped.
Add checks for "pcap_dump_flush()", and only enable the "-U" flag if
it's present. Clean up the handling of the "getopt()" argument and the
usage message to get rid of the pile of #ifdefs.
Add documentation for the "-L" and "-y" flags.
Tweak the description of "-r" to properly format "-w" in the text.
Young <dyoung@ojctech.com>, with some minor changes by Jason R. Thorpe
<thorpej@netbsd.org>, and further changes by me to:
use "-y" rather than "-D" to set the link type ("-D" was already
taken);
use libpcap APIs to map between data link type names and values;
supply stub versions of missing-but-needed libpcap APIs.
Update Jason Thorpe's e-mail address (Zembu is going away, if it hasn't
done so already).
the libpcap that comes with some platforms doesn't define it.
Check for "pcap_debug" and "yydebug" in libpcap in the configure
scripts, so that whichever one is present (which might depend on whether
libpcap was built with standard YACC or Berkeley YACC/Bison), if any is
present, will be used by the "-Y" flag (if none is present, "-Y" won't
be supported).
From Neil T. Spring: fixes for many of those warnings:
addrtoname.c, configure.in: Linux needs netinet/ether.h for
ether_ntohost
print-*.c: change char *foo = "bar" to const char *foo = "bar"
to appease -Wwrite-strings; should affect no run-time behavior.
print-*.c: make some variables unsigned.
print-bgp.c: plen ('prefix len') is unsigned, no reason to
validate by comparing to zero.
print-cnfp.c, print-rx.c: use intoa, provided by addrtoname,
instead of inet_ntoa.
print-domain.c: unsigned int l; (l=foo()) < 0 is guaranteed to
be false, so check for (u_int)-1, which represents failure,
explicitly.
print-isakmp.c: complete initialization of attrmap objects.
print-lwres.c: "if(x); print foo;" seemed much more likely to be
intended to be "if(x) { print foo; }".
print-smb.c: complete initialization of some structures.
In addition, add some fixes for the signed vs. unsigned comparison
warnings:
extract.h: cast the result of the byte-extraction-and-combining,
as, at least for the 16-bit version, C's integral promotions
will turn "u_int16_t" into "int" if there are other "int"s
nearby.
print-*.c: make some more variables unsigned, or add casts to an
unsigned type of signed values known not to be negative, or add
casts to "int" of unsigned values known to fit in an "int", and
make other changes needed to handle the aforementioned variables
now being unsigned.
print-isakmp.c: clean up the handling of error/status indicators
in notify messages.
print-ppp.c: get rid of a check that an unsigned quantity is >=
0.
print-radius.c: clean up some of the bounds checking.
print-smb.c: extract the word count into a "u_int" to avoid the
aforementioned problems with C's integral promotions.
print-snmp.c: change a check that an unsigned variable is >= 0
to a check that it's != 0.
Also, fix some formats to use "%u" rather than "%d" for unsigned
quantities.
return a structure pointer. Check whether the C compiler can handle
inline functions that return a structure pointer, not whether they can
handle inline functions that return an int, as at least some versions of
autoconf's AC_C_INLINE do.
Here is a patch that addresses a few SSL-related issues noticed:
1. The "/usr" directory is not the best choice to start looking
for SSL libraries when cross-compiling. The patch adds
"/usr/${host_alias}" at the front. Actually the test is quite
bogus anyway -- there might be no libcrypto.a library at all
(but e.g. libcrypto.so), so a better approach would be trying to
link against -lcrypto and seeing if that works. First with no
additional options (it might be in the default compiler/linker's
search patch, like on sane systems), then with the -L<dir>
option.
2. The "cast.h" and "rc5.h" headers should include the
"openssl/" path as that is what is used throughout the code.
Right now they are simply not found by configure.
3. The buggy CAST128 test should use a cache variable to permit
overriding by an educated user.
I think I may actually rewrite the test as described in #1 above
one day, but my time is quite limited and tcpdump is not my
priority task, so it might not happen soon. I won't mind if
someone does it earlier.
DECnet support currently assumes certain conditions instead of
checking for them explicitly. The following code checks if
dnet_htoa() is available, possibly in libdnet and also verifies
there is no declaration for the function in <netdnet/dnetdb.h>
before it decides to provide a substitute.
